Investigation of Sinkhole Formation with Human Influence: A Case Study from Wink Sink in Winkler County, Texas

The formation of sinkholes in Winkler County, Texas is concerning due to the amount of oil and gas infrastructure and the potential for catastrophic losses. Evidences of new and potential sinkholes have been documented, and determining the cause of these sinkholes is paramount to mitigate the devastating consequences thereof. Studies shown that the Wink sinkholes result from both natural and anthropogenic influences. Data depicting land-cover changes, alterations in the hydrologic systems, climate changes, and oil and gas activity were analyzed in an effort to better understand the link between these processes and sinkhole formation. Results indicate that the combination of these processes lead to the current state; Land cover changes were highest in shrub versus grasses, undeveloped to developed and croplands. Rises in temperature and a decrease in precipitation indicate a shift towards a more arid climate. Changes to the hydraulic system are a direct result of these land cover changes while the groundwater quality depict an environment prone to dissolution. Historical oil and gas activities have created pathways of meteoric water infiltration to the underlying limestone and evaporite formation. The combination of these processes create an environment that accelerates sinkhole formations. Understanding these processes allows for the development and implementation of better land practices, better groundwater protections, and the need for monitoring and maintaining aging oil and gas infrastructure.


Introduction
Sinkholes occur where overburden on the surface collapses into karst or a void within the subsurface. The formation of sinkholes is usually unforeseen and can create catastrophic losses of life and property, hence the term "hidden threat" is designed to describe this phenomenon [1,2]. Numerous sinkholes have been documented throughout Oklahoma is home to numerous sinkholes due to mining activities prominent in the area, and thus is unsuitable for human habitation. On February 23, 1998, an underground pipe rupture caused the development of a sinkhole by Interstate 15 [3].
Sinkholes are most likely to form in karst terrain and are often found in areas of artesian flow [4]; all 50 states have such terrain and 18% of this terrain is underlain with soluble lithologies [5], making the prediction of sinkhole formation almost impossible.
The study area contains many of the features, most notable of such features are the presence of limestone and dolomite formations, which are notorious for karst development, as well as underlying evaporates, which are highly soluble and prone to dissolution with the introduction of waters undersaturated with respect to the dominate salts such as Sodium chloride (NaCl) and Potassium chloride (KCl) among others. 3 Naturally occurring and anthropogenic pathways of water infiltration dominate the study area, making the development of sinkholes an issue of concern.
The development of Wink Sink #1 in 1980 illustrates this point. On June 3, 1980, the first sinkhole appeared in Winkler County, Texas. Within 24 hours, the sinkhole had grown to a width of 110 m [6]. The Wink Sink #1 was joined by Wink Sink #2 on May 21,2002, approximately 1,500 m to the south of the original sinkhole [1]. Neither collapse was predicted or expected. Luckily, there was no loss of life during the formation of either sinkhole. However, it has been proposed that humans played a significant role in the formation of both sinkholes, specifically through oil and gas development. Concerns over additional sinkhole development, and ways to predict them, have become a primary research topic within this area.
Several studies have investigated at the impact of early oil and gas activities in the Hendrick Field, which surrounds the Wink sinkholes along with sinkhole formation mechanisms [1,[6][7][8][9][10][11]  where the salt beds are overlain with other strata [12], as is the case within the study area.
The dissolution of salt and other evaporites, suffusion, and fracturing [11,13] of underlaying formations weakens overlying formations structurally. When the structure has been weakened past the point of supporting itself, it will collapse. This process continues until reaching the surface, forming the sinkhole.
Accounting for both the natural and anthropogenic sources of sinkhole formation is vital to understanding sinkhole formation and expansion. This research investigates environmental changes in conjunction with human activities with the goal of correlating the two in relation to the Wink Sink area. Historical changes in land use, precipitation and temperature were analyzed to develop a comprehensive explanation behind the 4 continual expansion of the existing Wink Sinks and currently developing sinkholes in the surrounding area. The aim of this study is to help local industries and governmental entities to make informed decisions regarding safe practices, policies and procedures when operating within the area to keep the public safe, and possibly prevent any future loss of infrastructure. Land cover is predominantly grasses, shrubs, and brush, followed by barren lands then developed areas and crops. This area is also characterized by high volumes of oil and gas

Methods
In order to better understand the link between anthropogenic and natural influences on sinkhole formation, changes in land cover and climate were analyzed. Data acquired to complete the historical assessment of land cover and land changes was obtained through Texas Natural Resource Information System (TNRIS) and U.S.
Geological Survey (USGS), the USDA Lab in Temple, TX and USGS. These datasets included layers that were added into ArcGIS to build maps spanning a 20-year period. 6 The type of changes assessed were water, developed land, barren land, brush land, grass land and crop land. Precipitation and temperature data between 1980 to 2015 were gathered from the National Oceanic and Atmospheric Administration (NOAA).
Alterations within these data sets were analyzed and interpreted as indications of overall climate change within the study area.
Data for water quality were obtained from the Texas Water Development Board The correlation of these datasets indicates a highly nuanced interrelationship between each set of data and sinkhole formation.

Land-cover Changes
Data from TNRIS indicates significant alterations in land cover between 1992, as shown in Figure2, and 2011, as shown in Figure 3. The most significant changes are the decrease of grass and barren lands, and the increase in cropland and developed areas.
These changes are summarized in Table 1. These changes influence the local hydrology by altering the elements in the local hydraulic system, most importantly run-off, infiltration and groundwater recharge. Land-cover changes can be traced to both 7 anthropogenic sources, including oil and gas developments, and naturally occurring sources, such as climate change.   Data from 1992-2011 depict a significant increase (approximately 63%) in developed land area. Urbanization increases the amount of runoff due to impervious areas with a decrease in evaporation, transpiration and infiltration within the developing areas [19]. Within the study area, developed land is not limited to an increase in urbanization.
Oil and gas activities increase the amount of impervious land cover due to the construction of lease roads, drilling pads and other infrastructures with little to no restoration of natural ground cover at the completion of these activities. The data also indicates an increase of cropland by approximately 173%. Land uses of pasture and cropland result in an increase in soil compaction and soil densities, which in turn resulted in precipitation runoff of 54.5% [20]. This alteration greatly increases the amount of runoff, while significantly impacting the local groundwater systems.
The loss of grasslands and increase of shrubs are attributed to the relationship between root systems, soil type, and water availability. While grasses develop deep root systems that are capable of deep soil moisture uptake, shrubs are capable of an even deeper penetration, developing root systems into partially weathered bedrock [20].
Additionally, a higher content of calcium carbonate within the soil may inhibit microbial activity, effectively limiting the amount of macronutrients available within the soils needed for grasses to thrive [21]. The overwhelming majority of soil within the study area is dominated by calcium carbonate-based lithology due to the deposition and erosion of vast limestone deposits throughout geologic history, the most recent during the Cretaceous period. Within the areas of shrub-based vegetation, eolian sediments (including minerals and organic matter) are trapped and settled within the zones of shrubs. This creates an environment that allows the shrub vegetation to thrive and overwhelm even more grassland. Grassland are also more sensitive to climate conditions 9 such as drought, while the shrub cover remains and thrives [21]. This change in vegetation leads to more surface run-off and less infiltration of precipitation.

Climate Changes
Analysis of precipitation and temperature data from 1980 to 2015 revealed an increase in annual average temperature with a decrease in annual precipitation indicating a shift to a more arid climate and drought conditions. The shift to more arid and droughtprone conditions alters both vegetation and soil conditions. During drought conditions, vegetation dies, thereby reducing groundcover, soil moisture uptake and evapotranspiration. The scarcity of precipitation decreases water supplies and increases the drying of surfaces soils [22]. Figure 4A is a linear regression of annual temperature data showing the average increase of annual temperatures through the past three decades, while Figure 4B plots the anomalies of annual temperatures. Figure 4C depicts the decline in precipitation rates and Figure 4D shows the anomalies within the precipitation data. This analysis confirms a shift to a more arid climate within the study area.

Groundwater Changes
The increase in runoff due to land cover changes combined with climate alterations and its subsequent effects on soils poses a unique issue within the study area.
With the anthropogenic alterations in ground cover limiting the area of soils capable of infiltration, the potential for the development of localized, new catchment basins and runoff pathways increases [2]. The concentrated influx of groundwater in these specific locations may potentially alter groundwater flow patterns and recharge zones. The increased burden of water in overlying strata coupled with increased withdrawal and subsequent lowering of hydraulic head in underlying aquifers, such as the Rustler, increases site-specific inter-aquifer flow and plays an active role in dissolution [2]. When overlying aquifers of higher hydraulic head were connected to underlying aquifers of lower head, water will flow into the lower-head aquifer through permeable strata or through improperly sealed boreholes [6].

Recent increases in production of groundwater from the Rustler Formation in the
Delaware Basin may be increasing this exchange. Typical production rates from Rustler wells are in the range of 300 gallons per minute (gpm) to over 700 gpm. There have been few studies on the impact this withdrawal has had on the Rustler and its hydraulically connected counterparts [23,24]. The Rustler, along with the Salado and Castille, has undergone several instances of deformation resulting in faulting, jointing and collapse throughout the Delaware Basin-all of which allow for the flow of water through the formation [25]. This allows for the introduction of meteoric waters into underlying evaporites and an increase in dissolution.
The alterations in land cover by creating new catchment zones and limiting recharge areas accelerates the process. Dissolution has been ongoing well into the Tertiary [2] and has continued to today. When water falls through the atmosphere and percolates through the over burden, CO2 is dissolved, forming weak carbonic acid (H2CO3). As the H2CO3 infiltrates the underlying carbonate and evaporite formation, the formation dissolves. The amount of dissolution can be estimated through TDS concentrations. As the amount of carbonate and evaporite dissolved in the water increases, TDS concentrations also increase. This dissolution occurs especially fast around the areas of focused infiltration [13] and dissolution rates within these evaporite layers can be up to three times faster than that are seen in limestone lithology [12], especially with the introduction of water undersaturated with respect to the dominate salt within the evaporite. As more waters percolate down and through the underlying Salado and Castille, the development of sinkholes could greatly increase in both the rate and quantity of the development.
12 Table 2. Aquifer properties in the study area The PVA is an unconfined aquifer characterized by unconsolidated alluvial and eolian sediments consisting of interbedded sand, silt, clay, caliche and gravels. Average  Figure 5B). Throughout the eight decades of water quality data available, the PVA bicarbonate ( Figure 5C) and pH levels ( Figure 5D Water quality parameters from these comingled areas were not analyzed in this study.
The depth of the Dockum ranges from 39.3 m to 396 m [28]. It produces fresh to saline waters and is used for agricultural and domestic purposes along with oil and gas supply [16] within the study area. The Capitan Reef aquifer is a fractured and karsted limestone aquifer. Depth to aquifer ranges from 914 m to over 1,219 m. Despite the use of numerous wells within the study area for oil and gas, water quality data was limited to one well over two decades.
The Capitan exhibited the poorest water quality within the study area. The poor water quality is a factor of formational age, residence time [31], depositional environment, depth, and oil and gas development [32][33][34]. The Capitan can be considered as one of the sources of the deep basin brines within the study area. Figure 5A shows TDS values two to three times higher than the PVA peak in 1950, ranging from 7,327 mg/L in 1990 to 4,845 mg/L in 2000. Figure 5B shows that chloride concentrations were 2,402 mg/L in 1990 and 1,070 mg/L in 2000. Bicarbonate values ( Figure 5C) were also nearly double in PVA and Dockum concentrations, yet pH values ( Figure 5D) indicate that the environment was more acidic. In 1990, pH was 6.41 and 6.59 in 2000. Due to the substantially higher quality water of the PVA and Dockum aquifers along with the presence of the semi-confining and confining lithology overlying the Capitan Reef complex, it has been inferred that there is little to no naturally occurring hydraulic connectivity between aquifers within the study area and any subsequent mixing or upwelling is a direct result of anthropogenic influences [35].

Oil and Gas Development
As the surrounding infrastructure of the Hendrick Oil Field ages, concerns over what are now seen as improper practices and their unforeseen consequences rise. Wells were drilled vertically, a mere 201 m apart from other wells. Numerous boreholes were deviated, and as was the case in Hendrick Well 10-A (the location of Wink Sink #1), explosives were used to allow for re-drilling of a straighter hole. Explosives were also used to increase production within oil-bearing zones [7], as a precursor to modern fracking techniques. The fracturing of rocks with explosives resulted in uncontrolled fractures being formed throughout the zone, unlike today's highly engineered and scientific methods. These fractures increased permeability for oil production, but also created the necessary conduit to allow the unsaturated waters to have continuous access to the dissolution-prone, underlying formations [2].
Records of plugging and abandoning wells are sparse, and regulations were drastically different than today. Improperly plugged wells, again, such as Hendrick Well 10-A [5], have created new pathways of water migration through the overburden, and there is insufficient data to determine where these wells are located or how many there are. As a pipe ages, it corrodes and the well itself can become a direct pathway to the delicate underlying strata. Wells that did not utilize secondary and tertiary casing and cementing are also potential pathways. If the well bore is lacking in cement throughout the entire borehole above the targeted zone, meteoric waters will use these pathways to flow. If cement was placed in the borehole behind the pipe, incomplete cementing jobs would still allow for water to flow behind the cement. Unbeknownst to the workers of the Hendrick Field, the actions taken then may have set in motion for the events that are culminating in the development of sinkholes today and for an unforeseeable time in the future.

Conclusions
The results of this study indicate sinkhole development in the Wink area of Winkler County, Texas is due to a combination of anthropogenic and naturally occurring circumstances. As the climate continues to change, the affects will be seen in the alterations of natural vegetation and soil behaviors. When this is coupled with anthropogenic alterations in land use and land cover, the development of sinkholes is accelerated. In order to protect the public and existing infrastructure, a more in-depth understanding of the relationship between the two is necessary. The results of this research have indicated that the historical and current practices of the oil and gas industry are one of the primary anthropogenic sources of sink hole formation within the study area. A more in-depth study on methods to contain, eliminate or mitigate the effect of aging oil and gas infrastructure coupled with better land cover and land use practices that include an awareness of runoff and drainage issues, management of local water table conditions and utilization of current technologies during urbanization within the study area could potentially allow for better prediction and containment of sink holes.
While the prediction will heavily depend on advanced technologies and consistent monitoring of known and potential sinkhole formation sites, re-stabilization efforts of the surrounding land cover and identification of failing infrastructure will be two key areas to consider to prevent the development of more sinkholes. The reintroduction of native grasslands, monitoring of both overlying and underlying aquifers, and proper drainage management are some of the available methods that this study suggests the possible stabilization of surrounding areas. Conducting site specific experiments with these goals would be the next step according to the findings within this study. While it may be beyond the scope of these experiments to alter the climate, altering the stewardship of the land is not beyond current scientific capabilities.

Author Contributions
S.E. collected detailed information on the sinkhole, calculated the environmental and groundwater changes, and analyzed the results with conclusions. J.H. designed the structure, developed the arguments, and contributed for the overall paper. All authors reviewed and approved of the final manuscript.

Funding
This work was supported by a Rising STAR (Science and Technology Acquisition and Retention) Program from the University of Texas System.